USENIX Security2025Top-tier venue
SEAF: Secure Evaluation on Activation Functions with Dynamic Precision for Secure Two-Party Inference
Abstract
Secure evaluation of non-linear functions is one of the most expensive operations in secure two-party computation, particularly for activation functions in privacy preserving machine learning (PPML). This work introduces SEAF, a novel framework for efficient Secure Evaluation on Activation Functions. SEAF is based on the linear approximation approach, but enhances it by introducing two key innovations: Trun-Eq based interval test protocols and linear approximation with dynamic precision, which have the potential for broader applicability. Furthermore, we classify common activation functions into several categories, and present specialized methods to evaluate them using our enhanced techniques. Our implementation of SEAF demonstrates 3.5× to 5.9× speedup on activation functions Tanh and Sigmoid compared to SirNN (S&P'21). When applied on GELU, SEAF outperforms Iron (NeurIPS'22) by more than 10× and Bolt (S&P'24) by up to 3.4×. For end-toend secure inference on BERT, the original GELU accounts for 31.3% and 22.5% of the total runtime in Iron and Bolt, respectively. In contrast, our optimized GELU reduces these proportions to 4.3% and 9.8%, eliminating GELU as a bottleneck in secure inference.
Ask Lune
Know this paper in minutes, not an afternoon.
Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.
1 top-tier paper has cited it since. Know what they found before your reviewers do.
Lune can draw on
the full paper · the top-tier papers that cite it · related work
Free to start. No credit card required.
Lineage
- CrypTFlow2: Practical 2-Party Secure Inference cites Oblivious Neural Network Predictions via MiniONN Transformations
- CrypTFlow2: Practical 2-Party Secure Inference cites SecureML: A System for Scalable Privacy-Preserving Machine Learning
- Oblivious Neural Network Predictions via MiniONN Transformations cites SecureML: A System for Scalable Privacy-Preserving Machine Learning
- ABY2.0: Improved Mixed-Protocol Secure Two-Party Computation cites Efficient Two-Round OT Extension and Silent Non-Interactive Secure Computation
- ABY2.0: Improved Mixed-Protocol Secure Two-Party Computation cites SecureML: A System for Scalable Privacy-Preserving Machine Learning
- Efficient and High-Accuracy Secure Two-Party Protocols for a Class of Functions with Real-number Inputs cites CrypTFlow2: Practical 2-Party Secure Inference
- Efficient and High-Accuracy Secure Two-Party Protocols for a Class of Functions with Real-number Inputs cites Oblivious Neural Network Predictions via MiniONN Transformations
- Efficient and High-Accuracy Secure Two-Party Protocols for a Class of Functions with Real-number Inputs cites SecureML: A System for Scalable Privacy-Preserving Machine Learning
Cited by top-tier papers1
Builds on11
- SecureML: A System for Scalable Privacy-Preserving Machine LearningPayman Mohassel, Yupeng ZhangS&P 2017 · 2,107 citations
- Oblivious Neural Network Predictions via MiniONN TransformationsJian Liu, Mika Juuti, Yao Lu, N. AsokanCCS 2017 · 800 citations
- ABY2.0: Improved Mixed-Protocol Secure Two-Party ComputationArpita Patra, Thomas Schneider, Ajith Suresh, Hossein YalameUSENIX Security 2021 · 307 citations
- CrypTFlow2: Practical 2-Party Secure InferenceDeevashwer Rathee, Mayank Rathee, Nishant Kumar, Nishanth Chandran et al.CCS 2020 · 294 citations
- Efficient Two-Round OT Extension and Silent Non-Interactive Secure ComputationElette Boyle, Geoffroy Couteau, Niv Gilboa, Yuval Ishai et al.CCS 2019 · 238 citations
Related papers
- Nimbus: Secure and Efficient Two-Party Inference for TransformersZhengyi Li, Kang Yang, Jin Tan, Wen-jie Lu et al.NeurIPS 2024 · 34 citations
- Iron: Private Inference on TransformersMeng Hao, Hongwei Li, Hanxiao Chen, Pengzhi Xing et al.NeurIPS 2022 · 209 citations
- SecFormer: Fast and Accurate Privacy-Preserving Inference for Transformer Models via SMPCJinglong Luo, Yehong Zhang, Zhuo Zhang, Jiaqi Zhang et al.ACL Findings 2024 · 23 citations
- Bicoptor: Two-round Secure Three-party Non-linear Computation without Preprocessing for Privacy-preserving Machine LearningLijing Zhou, Ziyu Wang, Hongrui Cui, Qingrui Song et al.S&P 2023
- PCFormer: Accelerating Privacy-preserving Transformer Inference by Partition and CombinationBo Zeng, Zhi Pang, Yuyang Zhang, Kai Zhao et al.AAAI 2026
